PromptLeak Scanner
Security test runner for AI applications — probes prompt injection, system prompt leakage, role-bypass, and unsafe-response patterns to find weaknesses before attackers do.
Loads an attack prompt library, runs each attack against a configured target (or against a captured response set), and analyzes responses for leaked system prompts, sensitive data, role-bypass behavior, and unsafe outputs. Outputs findings, risk-scored summary, and an engineer-friendly remediation report.
Catalog entry only — a full write-up lands closer to release.
Related across catalogs
- AutoPatch AISecurity project· Demo ready
Dependency vulnerability scanner and upgrade planner — scans package.json, requirements.txt, and Dockerfile base images, flags risky pins, produces PR-preview output.
- PhishGuard AISecurity project· Demo ready
Phishing email analyzer — parses .eml files, checks headers, link risk, sender spoofing, and language indicators, then explains findings in plain English.
- RepoSentinel AISecurity project· Demo ready
Repository security reviewer — scans a local clone for secrets, insecure configs, vulnerable dependencies, risky Dockerfiles, and surfaces PR-style fix suggestions.
- ScreenWall Security MonitorSecurity project· Demo ready
Security auditing tool for digital signage and kiosk fleets — audits signage URLs, kiosk configurations, public playlists, CSP, and browser-version risks across unattended screens.
- AttackReplay StudioSecurity project· Demo ready
Visual incident replay dashboard — turns safe sample logs into an attack timeline, attack path view, GeoIP map, and incident summary.
Want a heads-up when PromptLeak Scanner releases?